(360c) Enhancing Pathway Synthesis Using Genetic Information

The engineering of non-native synthesis pathways in microbial hosts has shown promise in the production of commercially useful biomolecules. In some cases, a choice for the synthesis pathway may be obvious. For example, there is only one known pathway for biosynthesis of 1,3-propanediol from glycerol. This pathway consists of two reactions, each catalyzed by a singular enzyme. More generally, the number of alternative pathways for a given product may be too large for experimental exploration. To facilitate the identification of possible synthesis pathways for a desired target biomolecule, current computational synthesis methods identify one or more possible synthesis pathway that can be added to the host. Each path consists of a series of non-native reactions that can be catalyzed by a number of possible genes. While there currently is a lack of data and consensus on the best pathway scoring methods, proposed metrics have been based on metabolic objectives such as yield, metabolic burden, and thermodynamic feasibility.

Despite progress, it is still difficult to identify the best synthesis pathway. Current synthesis methods assume uniform activity levels for all identified enzymatic reactions despite differences in kinetic parameters and solubility. To date, more than 1,000 prokaryotic genomes have been fully sequenced and annotated. Partial or draft genomes are available for more than 6,000 species. The total number of reactions listed in the Kyoto Encyclopedia of Genes and Genomes (KEGG) currently exceeds 8000. To exploit the diversity of metabolic enzymes across many different organisms, it becomes necessary for synthesis methods to efficiently compare the various genes available to catalyze a reaction along the identified synthesis pathway.

This work addresses the problem of identifying viable synthesis pathways and their gene implementation. In addition to identifying reaction steps required to construct each synthesis pathway, we identify the best organisms from which to cull the enzymes required to catalyze each reaction step. The major steps in our approach are as follows. First, ProbPath (Yousofshahi, M., K. Lee, and S. Hassoun, Metabolic Engineering, 2011) is utilized to construct non-native synthesis pathways for a desired product metabolite. ProbPath is graph-based pathway construction algorithm that utilizes probabilistic selection of reactions to construct pathways between a desired metabolite and a metabolite within the host. Second, synthesis pathways are ranked based on their yield using Flux Balance Analysis. Third, the solubility of each possible enzyme from each available organism is evaluated. To accomplish this, we create a database with reviewed protein sequences. The database contains information about specific activity for all proteins from specific organisms present in the BRENDA database. Our contribution is in creating a method for efficiently evaluating solubility and specificity for a large number of genes. We validate the workflow using several test cases.

The work described here advances the state of pathway synthesis by exploring the underlying gene design space, thus leading to more profitable design options when compared to using pathway synthesis tools that only explore the metabolic network design space. Importantly, advancing the design and engineering of synthetic biological systems will lead to reduced development cost, time, and effort, which in turn will enable new discoveries that have a positive impact on human health and the environment.